Code, AI, Glitch. Hand-crafting bugs since before AI could do it for me. Want the best AI tools + workflows + Vibe Coding? Join skool 馃憞
Arizona | Matrix
Joined April 2018
- Tweets16.2K
- Following428
- Followers2.6K
- Likes32.5K
Pinned Tweet
This openai/codex/chatgpt bug has been open on github since august
"I cant speak, because if I speak I am in big trouble..."
The US banned DeepSeek from government computers. Its own app stores your data in China, but the model is free to download, and its own page lists US companies that serve the exact same version. Same model, different mailbox. Full video on CodeGlitchTV.
The US Just Banned DeepSeek. Here's What It Means for You
youtu.be/GAZSXD6Bht0
Ofc I had to have Grok @bot "watch" this and create a detailed document on it xD
here's how i shipped 2,500 PRs last month to production
this was originally supposed to be for Cursor Compile in London. i couldn't make it since i was livestreaming for Grok @Bot Galaxy so i'm making it available for free here on X! watch it on 2x speed, i talk slowly
Grok bot is becoming great and helping me out more everyday:
- using confusing online platforms
- small research tasks
- daily security reports, morning routines and reporting
- doing other browsing automated tasks (several of them)
.....
Anyone else finding it more useful every day?
An open-source project lets Claude Code drive Blender, GIMP and LibreOffice by typing commands instead of clicking. 2,400+ tests pass, but its roadmap's benchmark for whether agents actually finish their tasks is still unticked. Click or type: full video on CodeGlitchTV.
Kimi K3.1 Pro is half Opus 5.5's price per token, but cached input, what an agent re-reads every step, costs $0.20 per million on both, and Anthropic says that's most of an agent's bill. Where it really saves money: full video on CodeGlitchTV.
Kimi K3.1 Pro vs Opus 5.5: Half the Price, Two Verdicts
youtu.be/8vYIF3Oz75M
Claude Code can take the ads off your TV's home screen and rebuild it into anything you ask for. In one test, the apps became eggs laid by a chicken in about five minutes. The catch is the door you open on your TV to let it in. Full video on CodeGlitchTV.
Claude Code Can Remove TV Home Screen Ads. Here's the Catch
youtu.be/jmtTIeUCngM
CodeGlitch retweeted
best tip for grok bot on mobile: talk to your bots in X Chat, not the grok bot app. the app is riddled with glitches, X Chat just works
on usage, 50x doesn't feel like 50x. coding or anything heavy still burns credits fast, so coding goes to claude code and grok bot gets the light stuff
anyone else running their bots from X Chat?
best tip for grok bot on mobile: talk to your bots in X Chat, not the grok bot app. the app is riddled with glitches, X Chat just works
on usage, 50x doesn't feel like 50x. coding or anything heavy still burns credits fast, so coding goes to claude code and grok bot gets the light stuff
anyone else running their bots from X Chat?
Tell a Claude project to run at most two threads at a time and it saves that as a preference, not a hard cap. New projects also run every thread on Opus at high effort, and threads carry on into your next usage window. The 5 ways Claude Code runs agents, on CodeGlitchTV.
Claude Projects Explained: 5 Ways Claude Code Runs Agents
youtu.be/Gj5jkWNDoIg
Does anyone still use X Pro? Or remember it?
This is the cleanest way for to keep up with X and my lists. Too bad Elon keeps burring it more and more, only for premium and above.
CodeGlitch retweeted
Anthropic names 3 things its new agent, Claude Duet, won't do even when you approve them: trade securities, move money between accounts, or access someone else's account. Messages, payments, signatures and deletions wait for your approval. Full video on CodeGlitchTV.
Anthropic names 3 things its new agent, Claude Duet, won't do even when you approve them: trade securities, move money between accounts, or access someone else's account. Messages, payments, signatures and deletions wait for your approval. Full video on CodeGlitchTV.
Claude Duet vs Grok Bot: The Agent That Asks First
youtu.be/H_TYJUclQh0
One pull request at Anthropic went through more than 50 rounds of AI review overnight. About 59% of what the reviewer kept flagging was code that was there from its first pass. The fix: a cap after 3 rounds, plus a few other changes. Full breakdown on CodeGlitchTV.
Claude Writes 90% of Claude Code. So Who Reviews It?
youtu.be/s0QFmM3PirU
